Towards a non-Gaussian model of redshift space distortions (2002.02683v2)

Abstract: To understand the nature of the accelerated expansion of the Universe, we need to combine constraints on the expansion rate and growth of structure. The growth rate is usually extracted from three dimensional galaxy maps by exploiting the effects of peculiar motions on galaxy clustering. However, theoretical models of the probability distribution function (PDF) of galaxy pairwise peculiar velocities are not accurate enough on small scales to reduce the error on theoretical predictions to the level required to match the precision expected for measurements from future surveys. Here, we improve the modelling of the pairwise velocity distribution by using the Skew-T PDF, which has nonzero skewness and kurtosis. Our model accurately reproduces the redshift-space multipoles (monopole, quadrupole and hexadecapole) predicted by N-body simulations, above scales of about $10\,h^{-1}{\rm Mpc}$. We illustrate how a Taylor expansion of the streaming model can reveal the contributions of the different moments to the clustering multipoles, which are independent of the shape of the velocity PDF. The Taylor expansion explains why the Gaussian streaming model works well in predicting the first two redshift-space multipoles, although the velocity PDF is non-Gaussian even on large scales. Indeed, any PDF with the correct first two moments would produce precise results for the monopole down to scales of about $10\,h^{-1}{\rm Mpc}$, and for the quadrupole down to about $30\,h^{-1}{\rm Mpc}$. An accurate model for the hexadecapole needs to include higher-order moments.